How a wearable with AI could be your social coach

MIT researchers are working on groundbreaking AI technology that could help those who struggle to read emotional and social queues.

ai and keeping human jobs primary

For those of us who have Asperger’s or who are just really bad at reading social cues and emotions, it can be challenging to engage with others and carry on an interesting conversation. And even if you are not, we have all had those moments when you cannot tell if someone is being sincere or not when he tells some outlandish story or statement.

To fix this problem, Mashable reported earlier this year that a pair of MIT researchers have developed a wearable “that could someday act as a real-time virtual social coach.” The wearable comes with artificial intelligence (AI), which can analyze a person’s speech patterns and vitals to determine what others are feeling.

Constructing a reader

The MIT researchers used a Samsung Simband, which can run custom algorithms. They created an AI which picks up the audio data from a speakers’ conversation and can even track some of their vitals such as heart rate and skin temperature.

But while the AI could pick up the data, the researchers had to work to teach the AI how to classify the conversation. They collected 30 conversations and trained two algorithms. One tracked whether an overall conversation was “happy” or “sad.” The other took five-second periods throughout the conversation and tracked whether it was “positive,” “negative” or “neutral.”

Through this, the AI system can keep constant track of a conversation and keep track of the speaker’s feelings. After a conversation, a user could go back through the conversation and get a better idea of what the speaker was likely thinking. The wearable would thus function as a social coach and thus teach someone with social anxiety social cues and improve their communication abilities.

Tuka Al Hanai co-authored a paper on the research team’s success and declared that “Developing technology that can take the pulse of human emotions has the potential to dramatically improve how we communicate with each other.”

Further work needed

Al Hani’s expectations seem certainly possible, especially since the new wearable can determine the overall tone of a story with about an 83 percent accuracy rate. But while this is impressive, the MIT researchers stressed that the system is still in its early stages. The hope is that the researchers can start using the system on the Apple Watch and improve its effectiveness by giving it more conversations to understand what is happy and sad. They would eventually release it with the goal of giving to those with Asperger’s or social anxiety.

Even if this system is complete, the privacy concerns of a wearable which is constantly recording others’ conversations should be obvious. The researchers did state that “a consumer version would obviously need clear protocols for getting consent from the people involved in the conversations.” But if that is the case, then the effectiveness of this system would be compromised by the obvious fact that we speak differently when being recorded.

Nevertheless, the development of this system could have applications beyond merely helping those with social anxiety. If an AI can learn to understand when a human conversation is “happy” or “sad,” then perhaps such functions could be used to improve AI-to-human communication as well as human to human communication.

This article is published as part of the IDG Contributor Network. Want to Join?

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.
Now read: Getting grounded in IoT